[Bugs] [Bug 1272450] New: Data Tiering:heat counters not getting reset and also internal ops seem to be heating the files

bugzilla at redhat.com bugzilla at redhat.com
Fri Oct 16 12:26:32 UTC 2015


            Bug ID: 1272450
           Summary: Data Tiering:heat counters not getting reset and also
                    internal ops seem to be heating the files
           Product: GlusterFS
           Version: 3.7.5
         Component: tiering
          Assignee: bugs at gluster.org
          Reporter: nchilaka at redhat.com
        QA Contact: bugs at gluster.org
                CC: bugs at gluster.org

Description of problem:
I observed that sometimes the heat of a file is not getting reset in the next
cycle. Also seems like internal operations like xattr changes are heating
files, which is not acceptable

Version-Release number of selected component (if applicable):

How reproducible:

Steps to Reproduce:
1.created a 2x2 vol and started it
2.attached a tier with pure-distribute of 4 bricks ->each disk with only 1GB
size(tried even with 2x2 hot layer)
3.Now, enabled ctr
4. Created a file f1 of size 700mb, which hashed to brick1 of hot tier
5. When idle the file got demoted
6. Now created f2 such that it got hashed to same brick as f1, ie brick1 of hot
tier and 700mb size
7. Waited for it to get demoted
8. Now i touched both f1 and f2 files to heat them, but as the space in hot
tier will be insufficient, I wanted to see the behavior
9. f1 got promoted , but f2 failed with tier log saying disk space not
sufficient which is perfectly valid
10. But the heat measure was still showing up in sqldb query.
11, Waited for f1 to get demoted, ie in the next cycle,But i saw that while f1
got demoted, f2 got promoted, as the heat counters were not reset.

Also there was newly read heat counter too seen.

Expected results:
>Heat counter should get reset
>Also, internal metadata read/writes or operations should not heat files
>read counters should not get set in this case

[root at zod glusterfs]# tail -f portugal-tier.log 
[2015-10-16 11:50:00.715894] E [MSGID: 109023]
[dht-rebalance.c:699:__dht_check_free_space] 0-portugal-tier-dht: data movement
attempted from node (portugal-cold-dht) to node (portugal-hot-dht) which does
not have required free space for (/lisbon.2)
[2015-10-16 11:50:00.716661] E [MSGID: 109037]
[tier.c:492:tier_migrate_using_query_file] 0-portugal-tier-dht: ERROR -28 in
current migration lisbon.2 /lisbon.2

[2015-10-16 11:50:00.716820] E [MSGID: 109037] [tier.c:1454:tier_start]
0-portugal-tier-dht: Promotion failed
[2015-10-16 11:52:00.728929] I [MSGID: 109038]
[tier.c:1008:tier_build_migration_qfile] 0-portugal-tier-dht: Failed to remove
[2015-10-16 11:52:00.734278] I [MSGID: 109038]
[tier.c:476:tier_migrate_using_query_file] 0-portugal-tier-dht: Tier 1
src_subvol portugal-cold-dht file lisbon.2
[2015-10-16 11:52:00.736125] I [dht-rebalance.c:1103:dht_migrate_file]
0-portugal-tier-dht: /lisbon.2: attempting to move from portugal-cold-dht to
[2015-10-16 11:52:22.250989] I [MSGID: 109022]
[dht-rebalance.c:1430:dht_migrate_file] 0-portugal-tier-dht: completed
migration of /lisbon.2 from subvolume portugal-cold-dht to portugal-hot-dht
[2015-10-16 12:17:16.194105] I [MSGID: 109028]
[dht-rebalance.c:3327:gf_defrag_status_get] 0-glusterfs: Rebalance is in
progress. Time taken is 83897.00 secs
[2015-10-16 12:17:16.194140] I [MSGID: 109028]
[dht-rebalance.c:3331:gf_defrag_status_get] 0-glusterfs: Files migrated: 15,
size: 0, lookups: 21, failures: 6, skipped: 0

Status of volume: portugal
Gluster process                             TCP Port  RDMA Port  Online  Pid
Hot Bricks:
Brick yarrow:/dummy/brick108/portugal_hot   49240     0          Y       30546
Brick zod:/dummy/brick108/portugal_hot      49240     0          Y       4923 
Brick yarrow:/dummy/brick107/portugal_hot   49239     0          Y       30524
Brick zod:/dummy/brick107/portugal_hot      49239     0          Y       4903 
Cold Bricks:
Brick zod:/rhs/brick1/portugal              49237     0          Y       2557 
Brick yarrow:/rhs/brick1/portugal           49237     0          Y       28413
Brick zod:/rhs/brick2/portugal              49238     0          Y       2575 
Brick yarrow:/rhs/brick2/portugal           49238     0          Y       28433
NFS Server on localhost                     2049      0          Y       11729
Self-heal Daemon on localhost               N/A       N/A        Y       11852
Quota Daemon on localhost                   N/A       N/A        Y       11748
NFS Server on yarrow                        2049      0          Y       32441
Self-heal Daemon on yarrow                  N/A       N/A        Y       32646
Quota Daemon on yarrow                      N/A       N/A        Y       32537

Task Status of Volume portugal
Task                 : Tier migration      
ID                   : 931de257-0dcd-4125-87a8-0cce35caca38
Status               : in progress         

[root at zod ~]# gluster v tier portugal status
Node                 Promoted files       Demoted files        Status           
---------            ---------            ---------            ---------        
localhost            7                    8                    in progress      
yarrow               0                    19                   in progress      
volume rebalance: portugal: success: 
[root at zod ~]#

You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
You are the assignee for the bug.

More information about the Bugs mailing list